fix: reduce reconnect backoff to 1s flat

Exponential backoff up to 300s made no sense with rotating Tor exits
where each reconnect gets a fresh IP. Single 1s delay is sufficient.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
user
2026-02-21 19:39:30 +01:00
parent 0064e52fee
commit 1ea72011b7
4 changed files with 7 additions and 13 deletions

View File

@@ -156,7 +156,7 @@ SASL EXTERNAL (cert + creds) > SASL PLAIN (creds) > NickServ IDENTIFY
## Reconnect Backoff
```
5s -> 10s -> 30s -> 60s -> 120s -> 300s (cap)
1s (flat, no escalation)
```
## PING Watchdog

View File

@@ -57,14 +57,8 @@ Once probation passes without incident:
On any disconnection, the bouncer reconnects with exponential backoff
(configurable via `backoff_steps`):
| Attempt | Default Delay |
|---------|---------------|
| 1 | 5s |
| 2 | 10s |
| 3 | 30s |
| 4 | 60s |
| 5 | 120s |
| 6+ | 300s |
Reconnection delay is **1 second** (flat, no escalation). Each attempt gets a
fresh random identity and potentially a different exit IP.
Each reconnection uses a fresh random identity.
@@ -321,7 +315,7 @@ captcha_poll_timeout = 120 # max seconds to wait for solve
# Connection tuning
probation_seconds = 45 # post-connect watch period for k-lines
backoff_steps = [5, 10, 30, 60, 120, 300] # reconnect delays
backoff_steps = [1] # reconnect delay (seconds)
nick_timeout = 10 # seconds to wait for nick change
rejoin_delay = 3 # seconds before rejoin after kick
http_timeout = 15 # per-request HTTP timeout